The overall aim of this study is to identify genes influencing coronary artery calcification (CAC), a direct measure of atherosclerotic burden, as well as endophenotypes for the atherosclerosis pathway. We use the large, multicenter, geographically diverse, epidemiologically defined, longitudinal, broadly and deeply phenotyped (on all major atherosclerosis pathway domains) NHLBI Family Heart Study-SubClinical Atherosclerosis Network (FHS-SCAN) data for a genome wide association scan (GWAS). We believe our family study allows optimization of several aspects of GWAS study design. To minimize Type I error, maintain statistical power, avoid stratification bias, incorporate linkage evidence, and to achieve lowest possible cost, we will employ a two-stage study design. In Stage 1, 1,000 unrelated FHS-SCAN Caucasian subjects (500 cases with highly calcified arterial lesions and an equal number of low CAC controls) will be genotyped using the Illumina 500K HumMap chip, consisting of gene-based haplotype-tagging single nucleotide polymorphisms (htSNPs). Unrelated case-controls are one of the most powerful designs for gene discovery, and power is important to offset the need to correct for so many multiple comparisons in a GWAS. But the main danger of using unrelateds is false-positive hits due to population stratification. The beauty of this two stage design, is that in Stage 2 those SNPs showing evidence for association in Stage 1 (adjusted for multiple comparisons) will be genotyped in the remaining FHS-SCAN sample of 2,767 subjects, which include family members of Stage 1 cases/controls. Family-based association analyses in Stage 2 will rule out false positives due to population stratification. Further, we can utilize our linkage results obtained on these families to augment the interpretation of the association results. This family design of the FHS-SCAN resource allows us to optimize the Stage 1 sample for its main purpose (power for discovery) by subselecting unrelated cases-controls, and to optimize the Stage 2 sample for its main purpose (validation/elimination of false positives) by utilizing entire families. This two stage approach has high power to detect any gene explaining at least 3-6% of a trait in the atherosclerosis pathway, through ht-SNPs that are at least within R2=0.80 of a functional variant. With the wealth of phenotypic characterization of FHS subjects in extended 3-generational families and the available linkage results, a genome wide association scan would allow rapid and efficient discovery of genes related to the development of atherosclerosis in humans. Such findings would be of great significance: they could enhance our understanding of the metabolic and mechanistic processes that lead to atherosclerosis and coronary endpoints and, thereby, suggest possible points of intervention or therapy. [unreadable] [unreadable] [unreadable]